Yea, I think I see what you are getting too.

But, I wanted to avoid 4 passes through the same records for each derived table. Isn't duplicating the FROM part of the query 4 times with different date ranges mean that SQL has to make 4 passes at the data or is it smart enough to do it all in one pass?

Can you do a small example?

BOb


>What about calculating the wages per rule per period and then aggregating them into counts? You could do it with either a view or derived table.
>
>-Mike
>
>
>>If it does, I don't see it.
>>
>>>Bob,
>>>
>>>I don't 100% understand what you're asking (might not be your wording, I'm really tired) but does the NULLIF() function help you here?
>>>
>>>-Mike
>>>
>>>>Ok Folks,
>>>>
>>>>Here a problem I need help with, I am sure you will all rise to the occasion as you usually do.
>>>>
>>>>I am trying to create a query that summerizes employee tax data by taxing entity. Each taxing entity has its own tax_rule_id. I am getting the data for our pay histories table, which has a master table py_ehmast and a child table of py_ehtax with all the tax records. Each mast record is a payroll transaction/check.
>>>>
>>>>I need to not only summarize the data for each tax_rule_id but I need the summary for an entered date range, then month-to date, quarter-to-date and year-to-date. The user enteres the date range and I determine the dates from that. Not a problem.
>>>>
>>>>I also need the count of employees that had taxable wages for the period specified. So, I have all this working, here is a simplified version of the query...
>>>>
>>>>

>>>>SELECT
>>>>tax_rule_id
>>>>,count(distinct case when chkdate between @pdbegin and @pdend then ee_id else null end) count_pd
>>>>,count(distinct case when chkdate between @mbegin and @mend then ee_id else null end) count_mtd
>>>>,count(distinct case when chkdate between @qbegin and @qend then ee_id else null end) count_qtd
>>>>,count(distinct ee_id) count_ytd
>>>>FROM py_ehtax t
>>>>JOIN py_ehmast m ON t.ehmast_id = m.ehmast_id
>>>>WHERE m.chkdate BETWEEN @ybegin AND @yend
>>>>

>>>>
>>>>Ok, all that works great. The problem is. We allow the user to do adjustments. What that does is put records with negative values into the py_eh???? files. So, the sum will come out to the correct amount.
>>>>
>>>>So, it should be possible that for an ee_id for a specic tax_rule_id the sum for any of the for time ranges could be zero. In that case, I don't want the ee_id included in the count.
>>>>
>>>>If you understand this, do you have any ideas how I could do this.
>>>>
>>>>I thought to add to the WHERE something like
>>>>
>>>>

>>>>ee_id IN (SELECT ee_id, tax_rule_id, sum(tax_cur)
>>>>FROM py_ehtax t1
>>>>join py_ehmast m1 on t1.ehmast_id = m1.ehmast_id
>>>>WHERE ee_id = m.ee_id
>>>>     AND tax_rule_id = t.tax_rule_id
>>>>GROUP by ee_id, tax_rule_id
>>>>HAVING sum(tax_cur) > 0)
>>>>

>>>>
>>>>But, that would only apply to the YTD part...
>>>>
>>>>Thanks,
>>>>BOb